Pneumococcal Conjugate Vaccination Schedules Acquisition, Immunogenicity and Pneumococcal Conjugate and Yellow Fever Vaccine Co-Administration Study

Background: Pneumococcal conjugate vaccines (PCV) effectively prevent pneumococcal disease but the global impact of pneumococcal vaccination is hampered by its cost. The evaluation of reduced dose schedules of PCV includes measurement of effects on immunogenicity and carriage acquisition compared to standard schedules. The relevance and feasibility of trials of reduced dose schedules is greatest in middle- and low-income countries, such as The Gambia, where the introduction of PCV resulted in good disease control but where transmission of vaccine-type pneumococci persists. We designed a large cluster-randomised eld trial of an alternative reduced dose schedule of PCV compared to the standard schedule, the PVS trial. We will also conduct a sub-study to evaluate the individual-level effect of the two schedules on carriage acquisition, immunogenicity, and co-administration of PCV with yellow fever vaccine, the PVS-AcqImm trial.

average of 35 annual cases of VT IPD among children aged 2-59 months. In 2016 we detected six cases of VT IPD and in 2017 we detected three. In 2016/17 we detected zero cases of VT IPD among children in the rst year of life. These data indicate that the introduction of PCV7/13 has now controlled VT IPD.
It is now evident that following the introduction of PCV13, herd protection has developed in The Gambia. The annual count of VT IPD in older children in PSP was six to ten before the introduction of PCV13 in 2011. Following the introduction of PCV13 the annual case counts in 2015, 2016, and 2017 were four, one, and zero respectively (author's own data). In the 5-14 year age group, IPD incidence declined by 69% (95%CI, -28%-91%) and radiological pneumonia by 27% (95%CI, -5%-49%) [11].
The prevalence of nasopharyngeal (NP) carriage of PCV13 VT in the BHDSS area before the introduction of vaccine was 47% in the under-5 year age group. In 2015 and 2017 the prevalence of vaccine-types was 17% and 15% respectively (author's own data). The downward trajectory of vaccine-type prevalence from 2009 to 2015 and 2017 suggests that the introduction of PCV13 has substantially reduced the prevalence of VT carriage. However, it is evident that transmission of VT pneumococci continues in the population.
The Gambian EPI schedule currently includes birth doses of BCG, hepatitis B and oral polio vaccine (OPV); visits at 2 and 3 months of age when OPV, pentavalent, rotavirus and PCV13 vaccines are scheduled; a visit at 4 months of age when OPV, injectable polio (IPV), pentavalent, rotavirus and PCV13 vaccines are scheduled; a visit at 9 months of age when measles-rubella and yellow fever vaccines are scheduled; conjugate meningococcal group A vaccine scheduled at 12 months of age was introduced in 2018; at 18 months of age OPV and measles-rubella vaccines are scheduled. Human papilloma virus vaccine delivered to school-age girls was introduced in 2020. In the event that polio is eradicated then OPV will be phased out and be replaced by IPV. The EPI is also considering the introduction of conjugate typhoid vaccine. Thus, in recent years the EPI schedule has introduced several additional injectable antigens and more may be added in the future.
Several studies indicate that immunological priming for an optimal PCV booster dose response may be more effective with fewer primary doses [12,13]. In addition, the immunological response to a booster dose following a single priming dose may reduce VT acquisition to a greater degree than following the standard 3+0 schedule [14]. As a result, a schedule with one primary dose and a later booster dose, that is a 1+1 schedule, may induce greater herd protection than the 3+0 schedule. A potential barrier to the use of booster doses of PCV in low-income countries is the lack of evidence of safe co-administration with YF vaccine, which in Africa is generally scheduled at 9 months of age.
This trial joins a global initiative to generate evidence about reduced dose schedules for PCV. WHO is engaged with this initiative having held a consultative meeting in February 2016. Studies investigating reduced dose schedules for PCV are underway in South Africa, Vietnam, India, and the UK. The UK introduced a 1+1 schedule nationwide in 2019. Our trial in The Gambia is critical to provide evidence from a typical African setting.

Rationale {6a}
In Fiji and The Gambia, one dose of PCV at 2 or 3 months of age signi cantly reduced carriage of VT pneumococci at 9 months of age [13,15]. A Dutch study showed that following two doses of PCV7 at the ages of 2 and 4 months, a booster dose at 11 months prevented VT carriage in the 2 nd year of life [14].
Data from a recent UK trial show that the immunogenicity of the PCV booster dose using a 1+1 schedule was equivalent to, or superior to, a 2+1 schedule for nine of the 13 serotypes in PCV13 [12]. Of importance to the Gambian setting, where serotypes 1 and 14 have been the most common serotypes causing IPD, IgG responses to those serotypes following the booster dose were superior in the 1+1 group. Almost all infants in both the 1+1 and 2+1 groups had IgG responses above the protective titre of 0.35µl/ml, for all serotypes except serotype 3, for which fewer reached protective thresholds in both schedules. Geometric mean IgG concentrations following the primary series were higher in the 2+1 schedule, although differing durations of time between vaccination and blood collection biased the results towards lower concentrations in the 1+1 group. The UK transitioned to a national 1+1 schedule in 2019 [16]. There is also suggestive clinical evidence that herd protection following the use of a 2+1 schedule is greater than with a 3+0 schedule [17].
The duration of protection of PCV is poorly de ned but the potential for greater antibody persistence following a booster dose compared to doses in early infancy [18,19] suggests that protection may be more long-lived when a booster dose is given [20]. Even though PCV has proven e cacious against serotype 1 in young African children, serotype 1 continues to cause epidemic meningitis in the African meningitis belt [21]. There is a strong rationale to test, in our current epidemiological setting, whether a 1+1 schedule will provide an overall non-inferior programmatic effect compared to the 3+0 schedule. It is important to note that as the time course after introduction of a vaccine extends, the direct effect of vaccination becomes less important and herd protection assumes an increasingly important role [22]. This trial aims to conduct an immunogenicity and acquisition sub-study within a larger pneumococcal vaccine schedule (PVS) eld trial. Immunogenicity data will be important to interpret the results of the eld trial. Immunogenicity data on the 1+1 schedule are needed in a typical African population given that administration of the rst dose is likely to be earlier than in developed countries, concentrations of maternally-derived antibody, which may affect responses to the primary dose(s), are different in populations with high pneumococcal transmission, and responses to PCV may be quantitatively different in these compared with other populations. Similarly, empiric measurement of the effect of the PCV13 booster dose on VT acquisition will provide direct evidence of the relative effect of the two schedules on herd protection and assist interpretation of the main PVS trial.
Demonstrating a superior immune response at 18 months of age and reduced VT acquisition following the booster dose will provide supporting evidence to the PVS eld trial assisting decision-makers in considerations of reduced dose schedules. Reducing the number of PCV doses in EPI schedules will impact on multiple elements of the challenges posed by this vaccine; reducing the costs to countries and GAVI, reducing the number of injections in schedules and providing greater exibility for the inclusion of other vaccines, reducing staff time and cold-chain requirements, and ultimately making EPI programmes more acceptable and sustainable. If countries can safely transition to 1+1 schedules the global uptake of PCV should accelerate with greater and more sustainable reductions in pneumococcal disease.

Objectives And Hypotheses
The primary objectives will be achieved by testing the following hypotheses: The serotype-speci c IgG concentration for VT at 18 months of age will be greater in 1+1 compared to 3+0 schedule recipients.
The rate of NP VT acquisition from 9-14 months of age will be reduced in 1+1 compared to 3+0 schedule recipients.
The proportion of participants with protective titres of yellow fever (YF) neutralizing antibodies will be non-inferior in those who receive co-administered compared to separately administered PCV and yellow fever vaccines.
The secondary objectives are to compare in alternative versus standard schedule groups, the: Rate of NP acquisition of non-VT pneumococci between 9-14 months of age.
Proportion of participants with NP VT colonisation at 6, 9 and 18 months of age.
Proportion of participants with pneumococcal VT IgG concentration ≥0.35µg/ml, 4 weeks after the primary series, 4 weeks after the booster dose at age 9 months, and at 18 months of age.
Pneumococcal VT opsonophagocytic antibody (OPA) titres following a single dose at age 6 weeks, following three primary doses, following the booster dose at age 9 months, and at 18 months of age.
Geometric mean concentrations (GMC) of pneumococcal VT IgGs 4 weeks after administration of PCV13 at 9 months of age with and without co-administration with YF vaccine.

Trial design {8}
The PVS acquisition/immunogenicity (PVS-AcqImm) is a parallel group, unmasked, cluster-randomised trial of the individual level effect (i.e. a population of individuals) of two different schedules of PCV13. This trial is nested within the PVS eld trial but designed for interpretation of effects at the individual level. This is a phase IV trial involving licenced products comparing alternative and standard schedules for PCV13 and separate versus co-administration with YF vaccine. We will test the superiority of the 1+1 compared to 3+0 schedule to reduce the rate of NP acquisition of VT pneumococci from 9-14 months of age and in terms of immune response at 18 months of age. We will test the non-inferiority of the immune response following separate and co-administration of PCV and YF vaccines. Approximately equal numbers of participants will be enrolled in each cluster allocated to the two groups in a 1:1 ratio.

Trial setting {9}
The PVS-AcqImm trial is being conducted in Upper River Region (URR) in the area covered by the BHDSS ( gure 1). The trial will be based at the Basse Field Station of MRCG at LSHTM. The BHDSS population is 178,510 (225 villages) with 19% of the population aged <5 years. The annual birth cohort is around 6,000. The area has a child mortality rate around 50 per 1,000 live births. There are 40 geographically separate clusters of villages assigned to attend geographically separate EPI clinics ( gure 1). These geographic clusters of villages were randomly assigned to the 1+1 or 3+0 schedule in the PVS eld trial. PVS-AcqImm participants will be selected from the 28 clusters closest to Basse town.

Participant selection
The sampling frame for selection to receive the interventions will be all infants resident in the 28 selected clusters. Residency will be de ned as: Born to or cared for by a parent or guardian who is resident for greater than 4 months as con rmed by BHDSS records or a household visit with the report of the household or compound head. Provisional residency may be established by the verbal report of the parent or guardian.
Born to or cared for by a parent or guardian who intends to be resident for greater than 4 months with veri cation at a household visit and report of the household or compound head. Provisional residency may be established by the verbal report of the parent or guardian.
The sampling frame will be developed as part of the BHDSS in the trial area. Pregnancies and births in all households are registered from: 4-monthly BHDSS enumerations of each household, village reporter records, and registration at EPI clinics. These data are electronically recorded in the eld and synchronized centrally on a weekly basis. A veri ed and updated sampling frame is available for use each week. The sampling frame lists the mother's name, infant's name, date of birth, father's name, name of household head, village name, compound number and individual ID number.
All resident infants are eligible for enrolment in the large PVS eld trial. Approximately 28 infants in each of the 28 clusters will be enrolled in the PVS-AcqImm trial with 784 participants for study of pneumococcal carriage and 336 for measurements of immunogenicity (table 2). Enrolment will proceed over a period of 4-10 months. A similar number of infants will be enrolled each month in each cluster selecting the rst presenting infant each month and each sequential infant until the monthly target is achieved and approximately 28 participants are enrolled in each cluster. If 28 infants cannot be enrolled in all 28 clusters in reasonable time, enrolments in 'slow to recruit' clusters may be distributed to other group-speci c clusters resulting in some clusters having less or more than 28 participants.

Eligibility {10}
Participants must meet all of the inclusion criteria and none of the exclusion criteria to be eligible.

Inclusion criteria
Resident in the study area.
Intention to reside in their residential cluster until 18 months of age.

Exclusion criteria
Intent to move out of the study area before 18 months of age.
Age greater than 10 weeks.

History of invasive bacterial infection or measles.
Receiving long-term antibiotic therapy, i.e. greater than 4 weeks.
HIV infection in the infant or mother.
Immunosuppressive therapy or immunode ciency disorder.
Contraindication to PCV13 -severe hypersensitivity to a previous dose.
Contraindication to YF vaccine.

Who will take informed consent? {26a}
Firstly, we informed the community leaders in the 28 selected clusters about the nature of the trial. Trained trial staff who speak the local languages presented the trial information and answered any questions.
Given that PVS-AcqImm enrolls participants also enrolled in the PVS eld trial, we obtain consent on one occasion for PVS and the follow-up and collection of specimens for PVS-AcqImm. As part of PVS procedures, we inform groups of parents or guardians of potentially eligible infants at immunisation clinics and other appropriate settings about the general nature of the PVS trial. Eligible infants making their rst presentation to the immunisation clinic after birth are identi ed. Trial staff determine the parent or guardian's rst language and literacy. More than three quarters of the adults in the trial area are not literate in English. Trained staff provide the trial information sheet to literate parents or guardians. If the parent or guardian is not literate, trial staff verbally explain standardised information concerning PVS and PVS-AcqImm as per the trial information sheet. Staff enquire whether there are any questions and seek informed consent. Mothers and guardians are encouraged to discuss participation with the infant's father before giving consent. Trial staff address the questions and concerns of parents or guardians.Trial staff who speak the language of the parent or guardian conduct the informed consent process.
If the parent or guardian is illiterate, an impartial witness is present during the informed consent process. Each impartial witness receives the information sheet and consent form. Impartial witnesses are reimbursed according to the standard operating procedures of MRCG at LSHTM. The consent of parents or guardians is recorded on a paper form. If literate in English the participant's parent or guardian signs and dates the consent form. If the parent or guardian is illiterate, the impartial witness attests to the participant's understanding, that informed consent is freely given, and the responses to the speci c questions on the consent form. The impartial witness marks the participant's responses to each of the speci c questions on the form. If the parent or guardian has understood the information, s/he thumbprint the consent form. The impartial witness signs the consent form and dates the participant's thumbprint. If a guardian provides consent, this is documented on the consent form and a statement of guardianship is obtained, with the signature of a witness if the guardian is not literate in English. The staff member obtaining consent records their name and signature on the consent form and provides an identical copy to the parent or guardian. The person obtaining consent also provides a copy of the information sheet to the parent or guardian (including the free-call contact details of two trial staff).

Consent for collection and use of participant data and biological specimens {26b}
Consent for the collection and use of participant data and biological specimens is speci ed in the trial information sheet and consent form. The consent form includes speci c con rmation, marked on the form and entered in the database, con rming consent for: the collection of speci c numbers and types of specimens, future research using the specimens, shipping of specimens overseas, and use of unidenti ed data via MRCG authorized data repositories.

Choice of comparators {6b}
Our choice to compare the 1+1 (doses of PCV scheduled at 6 weeks and 9 months of age) versus 3+0 (doses of PCV scheduled at 6, 10, 14 weeks of age) schedules for PCV vaccination is based on the hypotheses that the 1+1 schedule will provide su cient, but inferior, direct protection between 2 and 9 months of age, during which time the risk of VT disease in our setting is very low, and superior herd protection, and similar overall effectiveness compared to the 3+0 schedule. We also hypothesize that the 1+1 schedule will provide superior direct protection between 9 and 18 months of age compared to the 3+0 schedule. Finally, we chose to compare co-administration and separate administration of PCV and YF vaccines at 9 months of age as YF vaccine is scheduled at 9 months of age in most African countries.

Intervention description {11a)
The experimental intervention in this trial is the scheduling of PCV13 for infants resident in geographic clusters of villages to receive the rst dose due at 6 weeks of age and a booster dose scheduled at 9 months of age. The standard intervention is the scheduling of PCV13 for infants with doses due at 6, 10, and 14 weeks of age. A further intervention in the alternative schedule group is the scheduling in one group, of YF vaccine to be administered at 10 months of age and PCV13 at 9 months of age, and in another group co-administration of PCV13 and YF vaccines at 9 months of age. Following informed consent, infants are registered on the trial and a trial identi cation sticker is xed to the their infant welfare card. At each visit, trial staff identify the infant and the group allocation, based on the trial sticker and residential location. Trial staff record whether infants are migrating, or intend to migrate, within or out of the trial area.
PCV13 (Prevnar 13®) vaccine, manufactured by P zer Ltd, is licenced in many countries and has been approved for use in The Gambia since 2011. Stabilised YF vaccine is manufactured by Institut Pasteur in Dakar, with WHO prequali cation in 2001. The EPI procures these vaccines through global systems coordinated by UNICEF. This trial delivers PCV13 and YF vaccine in collaboration with, and through the structures of the Gambian EPI, and under the operational conditions of the public health system. The EPI receives PCV13 into a central cold-storage facility. Vaccine is transported to the regional centres either by a specially designed EPI 'cold van' or by cold storage units carried by Regional Health Directorate (RHD) vehicles. The RHD in URR is based in Basse. The URR RHD stores vaccine in solar refrigerators in Basse. From Basse, small volumes of vaccine are distributed on a monthly basis to ve 'base clinics' in Basse, Gambisara, Demba Kunda, Fatoto and Sabi. These ve base clinics administer vaccines in the 28 different geographic locations involved in PVS-AcqImm. Solar vaccine refrigerators are used for storage at the base clinics. The trial uses the existing EPI procedures to monitor and record the vaccine accountability and cold chain documentation with daily logs.
Immunisation is undertaken at four of the xed health centres in the study area on one or two days each week. Mobile clinics in the other 24 sites are held once or twice per month. Given the infrequency of EPI clinics, there is generally some delay in the time that vaccines are actually received. In 2016, coverage in the BHDSS of three doses of PCV at 12 months of age was 92% and coverage of one dose of measles vaccine was 82%.

Interventions -discontinuation or modi cation {11b}
Trial participants are discontinued from participation in the study if: Any clinically signi cant adverse event, intercurrent illness, or other medical condition or situation occurs such that continued participation in the study would not be in the best interest of the participant.
The parent or guardian so desires.
Participants who attend an EPI clinic outside their cluster and within the study area, but continue to reside in their original cluster, receive the trial schedule originally indicated on their infant welfare card and on the trial sticker. If participants migrate within the study area before completing their PCV schedule they continue to receive the trial schedule allocated in the cluster of their new residence. Participants who migrate after completing their PCV schedule do not receive any further doses of PCV in the cluster of their new residence. Parents in the 1+1 group are advised that if they migrate permanently out of the trial area they should attend the next available EPI clinic to complete the routine schedule for PCV. If an infant allocated to the 1+1 group visits an EPI clinic outside the study area the parent is instructed to request that their child receive the trial schedule indicated on the infant welfare card and study sticker. The study sticker includes free-call telephone numbers so that parents, or EPI staff outside the study area, may call for guidance. Infants resident in the alternative schedule clusters who decline consent are assigned to the national standard schedule.

Interventions -adherence {11c}
Adherence to the trial vaccination schedules is facilitated by exclusion criteria including an intention to migrate out of the study area in the next 18 months. Also, lists of infants allocated to the 3+0 group who have not completed three doses by the age of 5 months, and lists of infants allocated to the 1+1 group who have not received the booster dose by the age of 11 months, are generated every month to guide defaulter tracing at home visits throughout the study area.

Interventions -concomitant care and medications {11d}
PCV13 has been co-administered with Measles-Mumps-Rubella [23] and Measles-Mumps-Rubella-Varicella [24] vaccines but the results of both these studies do not report investigations of potential interference between the vaccines. PCV10 (Syn orix®) has been co-administered with YF vaccine in a study of an investigational GSK vaccine although results of investigations of potential interference between the vaccines have not been published. A different investigational PCV10 vaccine manufactured by the Serum Institute of India has been co-administered with YF vaccine with non-inferior immune responses [25]. Studies of YF vaccine co-administration with polysaccharide protein-conjugate quadrivalent meningococcal vaccine have not detected any adverse interaction [26]. The investigators are not aware of any data, or ongoing studies, that evaluate potential immune interference with the coadministration PCV13 and YF vaccine.
Participants admitted to hospital with an acute medical problem will have a blood culture taken, a plasma aliquot stored, a rapid malaria test done and haemoglobin concentration measured. Samples will not be collected from children admitted electively or those with surgical problems, trauma, acute burns or non-infectious neonatal problems. Participants admitted with suspected sepsis de ned according to standardised criteria will have a blood culture done and those with suspected meningitis will have a blood culture done and a lumbar puncture performed. For those admitted with clinical pneumonia, a danger sign, or focal chest signs, a blood culture and chest X-ray will be performed and pleural uid or lung aspirate obtained as clinically indicated. Other investigations will be done according to the clinical judgement of the attending clinician.

Post-trial care {30}
The trial may be stopped early if there is evidence that the risk of pneumococcal disease is greater in one compared to the other trial group. If the Data Monitoring Committee (DMC) recommends that the trial be stopped early, a joint meeting of the DMC, Trial Steering Committee (TSC) and Central Stakeholder Committee will make a recommendation to the Sponsor regarding post-trial procedures, including whether a dose of PCV be administered to children in a group found to be inferior. LSHTM carries clinical trial/non-negligent harm insurance and medical malpractice insurance applicable to this trial.

Outcomes {12}
The primary outcome of the immunogenicity study is serotype-speci c IgG for VT at 18 months of age, analysed at the ratio of the geometric mean concentrations (GMC) in 1+1 compared to 3+0 participants.
If serotype-speci c IgG concentrations at 18 months of age are greater in the 1+1 compared to 3+0 age group we will infer that the schedule including the PCV booster dose provides a superior immune response to the standard schedule at 18 months of age. This superior immune response may be associated with reduced acquisition of VT colonisation and reduced risk of VT disease in the second year of life.
The primary outcome of the acquisition study is the rate of NP VT acquisition from 9-14 months of age in 1+1 compared to 3+0 schedule participants. The rate of acquisition will be the number of VT pneumococcal carriage acquisitions detected in six monthly NP specimens collected between the ages of 9 and 14 months. If the rate of VT acquisition is reduced in 1+1 compared to 3+0 participants at this age we will conclude that the PCV booster dose provides superior protection from acquisition at this age with the implication of reduced transmission and potentially greater herd protection.
The primary outcome of the investigation of co-administration of PCV13 and YF vaccines is the proportion of participants with YF neutralizing antibody titres of ≥1:8 comparing those who receive the vaccines separately or co-administered. If the difference in proportions in the two groups is non-inferior we will conclude that there is no immune interference when the vaccines are co-administered. This nding would support a policy of co-administration of PCV13 and YF vaccine.
Secondary immunogenicity outcomes include comparison in alternative versus standard schedule groups, of: Pneumococcal VT opsonophagocytic activity (OPA) following a single dose at age 6 weeks, following three primary doses, following the booster dose at age 9 months, and at 18 months of age.
The proportion of participants with GMC of pneumococcal VT IgGs ≥0.35µg/ml, 4 weeks after the primary series, 4 weeks after the booster dose at age 9 months, and at 18 months of age.
Geometric mean concentrations of pneumococcal VT IgGs 4 weeks after administration of PCV13 at 9 months of age, with and without co-administration with YF vaccine.
These secondary immunogenicity outcomes will demonstrate the functionality of anti-pneumococcal IgG antibodies throughout the different schedules, which is required when investigating PCVs [27]. In addition, we will investigate the potential effect on PCV13 immunity of co-administration with YF vaccine, and so provide additional data on the safety of co-administration.
Secondary acquisition outcomes include the: Rate of NP acquisition of non-VT pneumococci between 9-14 months of age.
The proportion or participants with NP VT colonisation at 6, 9 and 18 months of age.
These secondary outcomes will provide information on the potential for the PCV13 booster dose to increase the rate of non-VT pneumococcal acquisition and the comparative effects of the two schedules on VT prevalence at different age points.

Participant timeline {13}
Participants in the trial are enrolled in one of three different groups with administration of intervention PCV13 and YF vaccine schedules as shown in table 2 which also shows the timing of specimen collection and measurement of endpoints. The SPIRIT gure ( gure 2) shows further details of the participant schedule for enrolment, interventions, and assessment.

Sample size {14}
For measurement of the effect of the PCV booster dose on the rate of acquisition of VT pneumococci, the smallest clinically important hazard ratio to detect is 0.75. A Dutch study of VT colonisation following a PCV booster dose at 12 months of age compared to no booster showed a relative risk reduction of 0.64 [14]. Thus, the hazard ratio of 0.75 is a more conservative estimate of the smallest relevant clinical effect size. Serotype-speci c anti-pneumococcal IgG to the serotypes included in PCV13 will be analyzed as a fold difference in GMCs as our interest is to test whether antibody concentrations are different and not necessarily correlates of protection. We de ne superiority as a 1.8-fold or greater difference in GMC for ≥10/13 serotypes, an approach used in other similar studies [28]. The 1.8-fold difference is based on data comparing IgG concentrations in Vietnamese children receiving PCV10 schedules of 3+1 or 3+0 in whom GMCs 1 month following the booster dose were 3-fold or more greater for all VT (pers. comm. P Licciardi 23 Apr 2018). Non-inferiority of response to YF vaccine will be de ned as the lower limit of the 95% con dence interval for the absolute difference in proportions with neutralizing titres of YF antibody ≥1.8 being greater than -10% [26,29].
The current prevalence of NP VT pneumococcal colonisation in those aged less than 2 years in the study area is 17% (author's own data). This measurement used selection of one or more morphologically different colonies, while the latex sweep serotyping method used in this study, is more sensitive to multiple serotype colonisation so we assume a slightly higher prevalence. The assumptions included in the sample size calculations are that the baseline risk of VT colonisation at any one point in time is 20%, participant withdrawal of 2%, death rate of 2%, loss to follow up of 3%, a 4% rate of cross-over between clusters, and 35% censoring of participants who do not experience VT colonisation. We have used local baseline data on the variability of cluster-speci c pneumococcal carriage to calculate an intra-cluster correlation of 0.02. The probability that VT pneumococci will be detected in the ve NP swabs collected between 10 and 14 months of age in the 3+0 group is 1-0.80 5 ; that is, the expected proportion of participants who acquire VT pneumococci is 0.67. Using a survival analysis logrank test and assuming a 5% level for statistical signi cance and 80% power, the study requires 784 participants with 28 participants in 14 clusters in each group to detect a hazard ratio of 0.75. The sample size was in ated from 392 to 532 with the intra-cluster correlation factor and then to 784 when taking into account rates of death, loss-to-follow-up, withdrawal, cross-over, and censoring due to no detection of vaccine-type colonisation.
The immunogenicity study includes two primary endpoints and the sample size calculation for each used α=0.025 and β=0.9. For the rst primary endpoint, our interest is in the relative magnitude of the IgG response in the two groups. Thus, we will test the alternative hypothesis of a ≥1.8-fold difference in GMCs between the 1+1 and 3+0 groups for ≥10/13 serotypes at 18 months of age. Baseline GMCs and their standard deviations for each serotype in the 3+0 group were taken from ndings in the USA [30] and Fiji [31]. Non-independence of observations within clusters was taken into account in the following way; given 112 participants in each group with 5% loss to follow-up (i.e. 106 participants per group) and eight participants in each of 14 clusters, and a variance in ation factor =1+(m-1)*ICC = 1 + (8-1)*0.01, the effective sample size in each group is n/VIF = 106/(1+(8-1)*0.01) = 99. Based on local cluster-wise variation in the mean prevalence of VT colonisation, the intra-cluster correlation = 0.01. Thus, simulations assumed 99 participants per group. Data were simulated so that there was a 1.8-fold difference in the IgG levels in the two groups, and differences in serotype-speci c GMCs were tested using a t-test. Withinsubject correlation between the multiple serotype-speci c endpoints was captured by a subject-level variation term with standard deviation 0.4 in a random-effects linear regression model. Superiority was declared for the overall conclusion if at least 10 out of 13 tests rejected the null-hypothesis in 5000 simulations. Table 3 shows the results of the power calculations indicating power for the overall conclusion of superiority is 99%.
For the second primary endpoint, power was calculated using a formula for cluster-randomised noninferiority study using Nquery + nTerim software. Calculations assumed that 95% of participants had YF neutralising antibody titres ≥1:8, [29] ICC = 0.01, and seven or eight participants per cluster (seven to account for loss to follow-up). Power to detect a 10% absolute difference in proportions with seven participants per cluster was 86% and 90% with eight participants per cluster. With the inclusion of 28 clusters and seven or eight participants per cluster, this leads to a sample size of 196 or 224 participants.

Recruitment {15}
All resident infants will be eligible for enrolment in the PVS eld trial, which enrolls approximately 150 participants per week. PVS participants may also be enrolled in PVS-AcqImm; 28 infants in each of the 28 clusters will be enrolled with a total of 784 participants for repeated measures of pneumococcal colonisation. Of the 392 participants resident in the 1+1 clusters, a sub-sample of 224 will be enrolled for measurement of immunogenicity endpoints, 16 in each of the 14 clusters. Of the 16 infants targeted in each 1+1 cluster a target of eight will be allocated to 1+1 with PCV13 -YF vaccine co-administration and a target of eight allocated to separate administration, i.e. 112 infants will be allocated to each of the co-administration and separate administration groups (table 2). Of the 392 participants resident in the 3+0 clusters, as subsample of 112 will be enrolled for the measurement of immunogenicity endpoints.
A similar number of infants will be enrolled each month in each cluster for approximately 10 months selecting the rst presenting infant each month and each sequential infant until the monthly target is achieved. For repeated measures of pneumococcal colonization, enrolment will be implemented by selecting the rst presenting infants in each cluster in each month until the target of 28 participants are enrolled in each cluster (less or more that 28 may be enrolled per cluster). In the 1+1 clusters, infants will be allocated to PCV13 -YF vaccine co-administration or separate administration groups according to a prepared random allocation list including 14 clusters with a target of 16 infants (less or more than 16 may be enrolled per cluster) in each cluster. In each of the 3+0 clusters, a target of eight infants will be enrolled in each cluster for measure measurement of immunogenicity endpoints, i.e. a total of 112 participants. If insu cient numbers of participants are enrolled in a cluster in one month, then additional participants may be enrolled in that cluster in the following month. If a participant withdraws or is lost to follow-up, additional participants may be enrolled in both groups to ensure su cient enrolment and the use of identical procedures in the two groups.

Sequence generation {16a}
Sixty-eight PVS trial clusters were randomised using a blocked scheme to ensure similar numbers of clusters were assigned to each group. Randomisation was strati ed by a binary variable correlated with 'high' or 'low' cluster-level incidence of clinical pneumonia. Randomisation was carried out in permutations using the above strati cation until selections achieved balance in terms of presence of a health centre and balance on population size between the two groups. In order to select the stratifying variable of cluster-level incidence of clinical pneumonia, cluster-level prevalence of 'high' or 'low' VT carriage in children with clinical pneumonia in the BHDSS was correlated with: clusterwise population density, rates of hospitalisation, clinical pneumonia, radiological pneumonia, IPD and mortality. Of the ve listed outcomes, clinical pneumonia incidence had the closest correlation with VT carriage prevalence, and thus was chosen as the stratifying variable.
Individuals selected for participation in PVS-AcqImm will be selected from the 28 clusters closest to Basse ( gure 1). Of these 28 clusters, 14 are allocated to each of the 1+1 and 3+0 groups, four of these 28 clusters include health facilities (two in the 1+1 group), and 14 are strati ed as high clinical pneumonia incidence (seven in the 1+1 group). Thus, the clusters selected for PVS-AcqImm are appropriately balanced for group allocation, clinical pneumonia incidence, and presence of a health facility.
Pre-prepared random assignment lists, by cluster, are used at the time of enroment to determine participant assignment to different schedules of blood collection. In the 1+1 clusters, the pre-prepared random assignment lists specify assignment to the PCV -YF vaccine co-administration or separate administration groups.

Concealment mechanism {16b}
A public event was held to announce group allocations of each cluster of villages in the PVS trial area.
Representatives of each cluster were present at the public event. Selection of the randomisation list at the public event involved random selection of one of 100 valid randomisation lists. Thus, the investigators and cluster representatives had no knowledge of the allocation sequence at the time of group allocation.

Implementation {16c}
An independent statistician prepared the cluster randomisation lists. Trial staff enroll the participants.
The trial data manager produced the lists for random assignment of participants to different schedules of blood collection and co-administration or separation administration of PCV -YF vaccines.
Who will be blinded {17a} Vaccinators and parents will be aware of the schedules used. Laboratory staff will be blinded as specimens will be labelled with a unique identi cation number that does not identify the study group.
Blinding of laboratory staff will avoid bias given the laboratory-based objectives of the study.

Procedure for unblinding {17b}
Given that participants are not blinded to their group allocation, a procedure for unblinding is not needed.

Assessment and collection of outcomes {18a}
Participants are under passive surveillance for clinical events as per the procedures of the PVS eld trial.
If participants present to a health facility in the trial area staff will provide standardized evaluation, investigation and recording of clinical safety events in an electronic medical record (see section 11d).
Details of the surveillance for clinical safety events are provided in an accompanying protocol paper for the PVS eld trial.
Selection bias is limited by pre-selection of infants within cluster to speci ed schedules of specimen collection. Baseline demographic and clinical information are recorded with a questionnaire administered at each visit documenting intercurrent illnesses and factors that may in uence pneumococcal colonisation. Data collection forms are not included in the protocol but are available on request.
The allowable window period for NP specimens collection is 14 days or more between swabs with the age 14 months specimen being collected no later than 15 months of age. Trained staff collect NP specimens according to WHO recommendations [32]. Biased case ascertainment is minimized by continuous quality control for the technique of NP specimen collection and allowing only a minimum number of staff collecting NP specimens. Flocked nylon swabs are placed in media and transported to the MRCG laboratory in Basse within 6 hours, or placed in a dry shipper for later transport to Basse.
NP specimens are processed in the Basse laboratory. A 10µl loop of vortexed NP specimen is inoculated onto blood agar with 5% gentamicin and incubated in CO 2 . Identi cation of pneumococcal colonies follows recommended methods [32]. Serotype-speci c anti-pneumococcal IgG and OPA will be measured in the Pneumococcal Laboratory at MCRI. A standardised ELISA will be used. Microtitre wells are coated with pneumococcal polysaccharide, depending on the serotype. To neutralize cell wall polysaccharide antibodies, diluted serum samples are incubated with cell wall polysaccharide and polysaccharide of serotype 22F, before further dilutions. A reference serum (007sp, Food and Drug Administration, Bethesda MD) is used and incubated with cell wall polysaccharide and 22F polysaccharide. Horse radish peroxidase conjugated anti-human IgG and the TMB Peroxidase Substrate system is used for detection. Three control sera are used to assess interassay variation. A standardised and validated OPA [34] will measure the serotype-speci c opsonophagocytic and killing activity of anti-pneumococcal IgG.
Neutralizing antibodies against YF virus will be assayed at BNITM. Neutralization titres will be expressed as the serum dilution yielding ≥50% neutralization (i.e., in which one or both of two duplicate standard cellular viral infections is blocked). If complete infection is observed at all serum dilutions, the neutralization titre is recorded as <1:4. Seroprotection is de ned as a neutralization titre ≥1:8.
An event of VT colonisation will be de ned as detection of a pneumococcus in a NP specimen belonging to serotypes 1, 3, 4, 5, 6A, 6B, 7F, 9V, 14, 18C, 19A, 19F, or 23F, using latex sweep methods. All other serotypes will be de ned as non-VT. Non-typeable isolates will be de ned as pneumococcal by colony morphology and biochemical means, not serotypeable by latex sweep or Quellung methods. An event of pneumococcal acquisition will be de ned as detection of a pneumococcal serotype in a NP specimen that was not detected in the previous NP specimen.

Participant retention and complete follow-up {18b}
Participants are only eligible for enrolment if intending to reside in the trial area until 18 months of age.
The informed consent process allows time for discussion among family members and we speci cally aim to discuss consent with the father of the child. We provide participants with a trial information sheet that includes free-call phone numbers that can be used at any time. Trial information speci c to every individual participants is provided in the form of a pictorial guide to the follow-up timeline and specimen collection schedule. Trial staff use the phone numbers of parents to communicate and facilitate complete follow-up. We provide instructions to participants if attending an EPI clinic outside the trial area that facilitate administration of vaccine according the the trial schedule.

Data management {19}
A data management plan has been prepared and approved by the MRCG Head of Data Management and is available on request. Data are collected on eCRFs using a standardised format. Electronic data capture (EDC) is done o ine using encrypted devices which is then synchronised with a central server on a weekly basis. Trial staff attend all EPI clinics, con rming the identity of all infants and record immunisation data in real-time. Trial staff generate source data on electronic devices at regular visits as per the trial timeline.
Data entered into eCRFs will be monitored for completeness and consistency against the relevant source documents. Independent trial monitors undertake 100% veri cation of source data for the primary endpoints and informed consent. Anomalies identi ed are reconciled with the source.
Front-end data quality checks are programmed into the data capture application. Backend edit checks and validation checks are built-in to monitor the validity of data (e.g. to identify inconsistent dates and times, and clinical and antropometric measurements outside de ned ranges, etc.). Data queries are generated weekly by the data manager for resolution by trial staff. Reports of data quality are generated periodically.
The trial database is housed on a secure network server with restricted access to the backend. The backend comprises a MSSQL database which will be regularly backed up as part of the organisation's disaster recovery plan. An in-house web based application is used for the database structure, using the PHP\ASP.Net platform connecting to the MSSQL backend database. A recovery point objective is set so that systems and data will be restored back to their prior status 24 hours prior to a failure. Daily and monthly backups of all media servers are done to achieve this objective.
The e-CRF is used as the speci cation for the design of the database. An annotated CRF indicates the relationship between the variable names in the database and the elds in the CRF. A data dictionary uses a standard template, include a list of database variable names, variable descriptions, data types, and sources, valid values and in-built edit checks. Standard data coding will be used (e.g. latest MedDRA version accessed through the internet will be used for adverse event reporting).
The long term storage of research records is done in accordance with MRCG policies and procedures for archiving. All paper records will be held for at least 10 years in the Unit's archiving facility. Data in electronic format will be held inde nitely on our Electronic Data Repository. Guideline for Good Clinical Practice E6 (R2 Step 4).

Con dentiality {27}
No subject identi able information (names, addresses etc) is entered into the database. Identi able information is stored in the electronic medical record, vaccination and BHDSS databases, encrypted and password-protected, and accessible to only a limited number of staff involved in the care of patients. Trial monitors only access pre-speci ed, non-identi able data. Informed consent documents are stored in locked, re resistant ling cabinets to which only the Principal Investigator and a limited number of delegated clinical trials personnel have access. Data are backed up at the end of each day in the eld on encrypted ash drives to prevent any data loss during transit. All computers within MRCG are access controlled with strong password policies that prevent unauthorized access to networked user machines. Users to whom network access has been given are granted necessary privileges to the trial database based on their trial roles.
Collection, laboratory evaluation and storage of biological specimens for genetic or molecular analysis

{33}
The trial collects blood and NP specimens but procedures do not include genetic or molecular analyses of human material.

Statistical methods for primary and secondary outcomes {20a}
The primary analysis of the effect of the PCV13 booster dose on pneumococcal acquisition will use data from the six NP specimens collected between 10 and 14 months of age. Detection of homologous serotypes within an individual on multiple occasions will not be counted as multiple acquisitions. That is, the maximum number of acquisitions for one participant will be ve. Analysis will employ a Cox proportional hazards regression model comparing the hazard ratio for VT acquisition in the 1+1 compared to 3+0 group including variables to adjust for previous serotype-speci c acquisition and trial cluster. The interpretation of the models will be at the individual level.
Secondary analyses of the endpoint of non-VT acquisition will use the same approach as described above. For the endpoint of VT prevalence at 6, 9, and 18 months of age we will compare proportions while also taking into account trial cluster.
Analyses of immunogenicity data will be per-protocol for non-inferiority endpoints and otherwise by modi ed intention-to-treat with all individuals included by randomised group if they have a laboratory result. For the primary endpoint, we will compare The VT serotype-speci c the GMCs in the two groups using linear regression including co-variates of baseline IgG, cluster, and season of PCV13 booster if associated with GMC, using separate models for each serotype. For calculation of GMCs, a logarithmic transformation will be applied to the IgG concentration value prior to analysis. The 1+1 schedule response will be deemed superior if the ratio of GMCs is ≥1.8 for ≥10/13 serotypes at the 5% level of signi cance.
Proportions of participants with YF neutralizing antibody titre ≥1:8 will be calculated with exact binomial 95% con dence intervals (CI). The CI for the difference in the proportions will be computed using the Miettinen-Nurminen method (or another appropriate method), including cluster and/or season as needed.
If the lower limit of the CI for the difference in proportions is greater than -10%, response in the coadministration group will be deemed non-inferior.

Interim analyses {21b}
There are no planned interim analyses.

Additional analyses {20b}
An exploratory analysis will include repeated acquisitions of homologous serotypes, that is counting repeated acquisitions separately, testing whether the rate of individual VT acquisitions, including repeated acquisitions, is reduced in the 1+1 compared to 3+0 group.
Analysis methods to handle non-adherence or missing data {20c} The protocol speci es enrolment of additional participants if a participant is lost to follow-up. Analysis will handle events of non-adherence by restricting the data included based on certain conditions. Receipt of the third dose of PCV13 in the 3+0 group at greater than 23 weeks of age will be considered outside the eligible window for per-protocol analysis. Receipt of the booster dose of PCV13 in the 1+1 group at greater than 11.5 months of age will be considered outside the eligible window for per-protocol analysis.
Receipt of YF vaccine in the 1+1 YF co-administration group at greater than 11.5 months of age will be considered outside the eligible window for per-protocol analysis. Receipt of YF vaccine at greater than 12.5 months of age by infants allocated to separate administration of PCV13 and YF vaccines will be considered outside the eligible window for per-protocol analysis. Inclusion of the age 18 months blood specimen in per-protocol analysis will require collection no later than 20 months of age. Analyses will be conducted with the available data.
Access to the protocol, participant level-data and statistical code {31c} The protocol is available on request. The data generated will be suitable for sharing in an anonymised format. The data will be in CDISC ODM format that is an internationally recognized standard. Data will not be deposited into a central repository but held securely by MRCG. The trial is registered with the International Standard Clinical Trial Registry Network (ISCTRN) to maximize its visibility to other interested parties. Summary data will be provided through ISCTRN at the end of the study. Data sharing will follow MRCG policy [36]. Applications for access to the complete datasets will need to be made to the MRCG Unit's Archives department who will then forward it to the Scienti c Coordinating Committee (SCC) of the Unit. All requests for the dataset will be reviewed by the SCC and also by The Gambia Government/MRC Joint Ethics Committee (GG/MRCG JEC) to establish that the request is in order to perform scienti cally appropriate analysis. The datasets collected within the trial will be available to other users once all relevant trial related publications in scienti c journals have been accepted. Prior to this point requests will be considered on a case by case basis.
For practical reasons this time period for trial related publications may be indicative and might need to be revised if delays occur. Different periods may be applied to different datasets, e.g. to take account complexity of cleaning and documentation.
Timing will depend on the trial's data collection patterns.
In relation to timing, the terms could, for instance, be expressed as follows: "6-months after the end of the current grant period", "12-months after new data collection to allow for data cleaning and documentation" or "3-months following the rst publication of ndings based on the data.
Statistical code will be available on request.
Coordinating centre and trial steering committee {5d} The trial management group meets every week, and includes the Principle Investigator, Trial

Data monitoring committee {21a}
The DMC is independent of the Sponsor and composed of a Chairperson, expert clinician, expert statistician, and an independent statistician. The role of the DMC is described in its Charter (see Supplentary material) and is to protect and serve trial participants and to assist and advise the PI so as to protect the validity and credibility of the trial. The DMC monitors the safety of the trial, reviews its progress and accruing data, makes recommendations to the TSC whether the trial should continue, be terminated, or modi ed, and determines if interim analyses should be undertaken. The DMC also considers data quality, recruitment, compliance with the protocol, sample size assumptions, data emerging from other related studies, requests for interim trial data, and the nal data and its interpretation. Materials and discussion during meetings are con dential. The investigators intend to publish the results of the trial in peer-reviewed scienti c journals. All trial publications will follow MRC guidelines for Open Access publishing. The results of the study will be disseminated to the parents/guardians of each participant and the communities in the study area. The ndings of the trial will be presented the Central Stakeholders Committee that includes representatives of of the Ministry of Health EPI and other regional and central health authorities. The results of the trial will be presented to the WHO EPI department and its Strategic Advisory Group of Experts on Immunisation.

Discussion
The start of recruitment in the trial was delayed in 2020 due to restrictions related to the COVID-19 pandemic. Since the beginning of recruitment we have not experienced any signi cant interruptions to trial implementation.

Trial status
The current protocol version is 5.0, dated 24

Role of sponsor and funder {5c}
The trial sponsor and funders are not involved in the study design; collection, management, analysis and interpretation of data; writing of the report; the decision to submit the report for publication, and will not have authority over any of these activities.   Schedule of enrolment, interventions, and assessments

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. ICDInfantPVSAcqImm11YFseparv4.0clean.docx